// Skill profile
scenario-labs/skills is a comprehensive creative-production skill suite that teaches AI coding agents to generate production-ready images, video, audio, textures, and 3D assets through the Scenario MCP server. The skills handle model routing (picking the right generative model for the task), cost estimation before spending, and character/brand consistency across outputs. Beyond the core generative skills, the repo includes expert DCC tools that drive Blender, Maya, and ZBrush installed on the user's machine, plus game engine integrations for Unreal Engine, Unity, Godot, and Roblox Studio. Install with `npx skills add scenario-labs/skills`. Works with Claude Code, Cursor, Codex, GitHub Copilot, and dozens of other agent harnesses. Actively maintained with regular releases (v0.47.0 as of Sep 2026). MIT licensed.
scenario-labs/skills is a skill suite from Scenario Labs that turns AI coding agents into creative production pipelines. Where most agent skills focus on code, documentation, or DevOps, this package addresses a different gap: generating and managing visual, audio, and 3D assets directly from an agent session. The skills route requests through the Scenario MCP server, which handles model selection, cost estimation, and output consistency — so the agent doesn’t need to know which generative model to call or how much it will cost.
The package splits into two layers: core generative skills and expert DCC/game-engine tools. The core skills cover image generation, video generation, audio (including two-speaker dialogue), texture creation, and 3D asset generation. The expert tools go further — they drive professional creative software (Blender, Maya, ZBrush) and game engines (Unreal Engine, Unity, Godot, Roblox Studio) installed on the user’s machine, turning the agent into an operator for these tools.
Model routing: The skills automatically select the right generative model for each task based on the prompt, desired output type, and quality requirements. Users don’t need to know which model to call — the skill handles it.
Cost estimation: Before any generation runs, the skill estimates the cost and presents it. No surprise bills from accidental high-resolution batch generations.
Character and brand consistency: The skills maintain visual consistency across outputs — if you generate a character in one image, subsequent generations preserve that character’s appearance. Brand assets (logos, color palettes, style guides) are respected across the pipeline.
DCC expert tools: Skills under skills/dcc/ drive Blender, Maya, and ZBrush. The agent can create scenes, manipulate geometry, set up materials, and render outputs using professional tools already installed on the user’s machine.
Game engine integrations: Skills under skills/game-engines/ cover Unreal Engine, Unity, Godot, and Roblox Studio. The agent can set up projects, import assets, configure materials, and build levels through these engines.
Game developers use the skills to generate concept art, textures, and 3D assets directly from their agent session, then push them into their engine of choice. Content creators use the video and audio skills for production workflows. Indie developers who can’t afford a full art team use the pipeline to generate consistent visual assets. Teams with established Blender or Maya workflows use the DCC tools to automate repetitive tasks.
The core generative skills route through the Scenario API, which has its own pricing — the cost estimation feature helps manage this but generation is not free. The DCC and game engine tools require the respective software to be installed locally. Quality of generated assets depends on the underlying models and prompt quality. Professional production workflows may still need human review and polish on generated outputs.
Game developers, 3D artists, content creators, and indie studios who want to integrate AI-powered asset generation into their agent workflows. Teams using Blender, Maya, Unreal Engine, or Unity who want to automate repetitive creative tasks. Anyone who needs production-ready visual, audio, or 3D content generated from natural language prompts.
AI agents that work well with Scenario Skills.
Anti-AI-slop design skill — 48 themes, task-driven layouts, browser-based slop lint via Puppeteer, and a 755-image photo library that force Claude Code, Cursor, and Codex to generate UI that looks intentionally designed, not template-stamped.
Agent skill that removes signs of AI-generated writing from text — rewrites agent output so it reads like a person wrote it, matching voice and preserving facts.
39 editorial diagram types for Claude Code, Codex, and Pi — self-contained HTML and SVG output with no shadows, no Mermaid slop, and brand-customizable styling.